Cross-sectional associations between carotid artery stenosis (CAS) on the one hand, and parameters of glycaemia and specific insulin levels on the other, were investigated in an age, sex, and glucose tolerance stratified random sample from a 50-74-year-old Caucasian population. Subjects treated with insulin or oral hypoglycaemic agents were classified as having known diabetes mellitus (KDM) (n = 66). Using two oral glucose tolerance tests, and based on the World Health Organisation criteria, all other participants were classified as having a normal (NGT) (n = 287), an impaired (IGT) (n = 169) or a diabetic (NDM) (n = 106) glucose tolerance. CAS was defined haemodynamically using duplex scanning. The crude prevalences of only moderate (16-49%) CAS were 6.6%, 7.1%, 5.7% and 12.1% in NGT, IGT, NDM and KDM subjects, respectively. For any severe (> or = 50%) CAS, crude prevalences were 2.8%, 4.7%, 9.4% and 7.6%. The prevalence of any severe CAS was higher in NDM (p < 0.01) and KDM subjects (p = 0.07) than in NGT subjects. The prevalence of a history of stroke or transient ischaemic attack was 1.7%, 1.8%, 2.8% and 1.5% in NGT, IGT, NDM and KDM, respectively. In univariate logistic regression analysis, HbA1c, serum fructosamine, fasting and 2-h post-load glucose were significantly associated with any severe CAS. In multivariate analyses controlling for other risk factors, only HbA1c and 2-h post-load plasma glucose remained significantly associated (odds ratios: 1.29 per % and 1.09 per mmol/l, respectively) in separate models. No association could be shown between either fasting or 2-h post-load specific insulin and any severe CAS in either univariate or multivariate analyses. In conclusion, HbA1c and 2-h post-load plasma glucose are independently associated with any severe CAS, whereas specific insulin is not.